Comparison of Reconstruction Algorithms for Decreasing the Exposure Dose During Digital Tomosynthesis for Arthroplasty: a Phantom Study

To explore the possibility of decreasing the radiation dose during digital tomosynthesis (DT) for arthroplasty, we compared the image qualities of several reconstruction algorithms, such as filtered back projection (FBP) and two iterative reconstruction (IR), methods maximum likelihood expectation maximization (MLEM) and the simultaneous iterative reconstruction technique (SIRT) under different radiation doses. The three algorithms were implemented using a DT system and experimentally evaluated by contrast-to-noise ratio (CNR), artifact spread function (ASF), and power spectrum measurements on a prosthesis phantom. The CNR and ASF data were statistically analyzed by a one-way analysis of variance. The effectiveness of each technique for enhancing the visibility of the prosthesis phantom was quantified by the CNR (reference dose vs. 20 % reduced dose in FBP, P = 0.62; reference vs. 37 % reduced dose in FBP, P = 0.16; reference vs. 55 % reduced dose in FBP, P < 0.05; reference vs. 20 % reduced dose in IR, P = 0.92; reference vs. 37 % reduced dose in IR, P = 0.40; reference vs. 55 % reduced dose in IR, P < 0.05) and ASF (reference dose vs. 20 % reduced dose in FBP, P = 0.25; reference vs. 37 and 55 % reduced dose in FBP, P < 0.05; reference vs. 20 % reduced dose in IR, P = 0.16; reference vs. 37 and 55 % reduced dose in IR, P < 0.05). The power spectra under the reference and reduced doses are equivalent. In this phantom study, the radiation dose of the reference dose could be decreased by 20 % with FBP and IR for consideration of common factors.